Method and system for measuring structure based on spectrum

Abstract

A method for measuring a structure based on a spectrum, includes obtaining a first model that includes a first sub-model and a second sub-model following the first sub-model and is trained based on simulation data, generating a second model including a third sub-model identical to the first sub-model, training the second model based on sample spectrum data generated by measuring spectra of sample structures, and estimating, based on the trained second model, the structure from measured spectrum data generated by measuring a spectrum of the structure.

1 . A method for measuring a structure based on a spectrum of the structure, the method comprising:
 obtaining a first model trained based on simulation data, the first model including a first sub-model and a second sub-model following the first sub-model;   generating a second model such that the second model includes a third sub-model generated from at least a portion of the first sub-model;   training the second model based on sample spectrum data generated by measuring spectra of sample structures; and   estimating, based on the trained second model, the structure from measured spectrum data generated by measuring the spectrum of the structure.   
     
     
         2 . The method of  claim 1 , wherein the obtaining the first model comprises
 generating virtual spectra by simulating virtual structures, and   wherein the simulation data represents the virtual structures and the virtual spectra.   
     
     
         3 . The method of  claim 2 , wherein the obtaining the first model comprises
 verifying the first model based on an error between output data of the first model and the simulation data.   
     
     
         4 . The method of  claim 1 , wherein
 the second model further comprises a fourth sub-model following the third sub-model, and   the training the second model comprises fixing the third sub-model such that the third sub-model is not trained while training the fourth sub-model based on the sample spectrum data.   
     
     
         5 . The method of  claim 4 , wherein
 each of the first sub-model and the third sub-model comprises a convolution network, and   each of the second sub-model and the fourth sub-model comprises a fully connected network.   
     
     
         6 . The method of  claim 1 , further comprising:
 verifying the second model based on an error between output data of the second model and measured structure data of the sample structures.   
     
     
         7 . The method of  claim 6  wherein the verifying the second model comprises
 extracting a first sample and a second sample from the simulation data, 
 obtaining first output data and second output data of the first model, the first output data and the second output data respectively corresponding to the first sample and the second sample, 
 obtaining third output data and fourth output data of the second model, the third output data and the fourth output data respectively corresponding to the first sample and the second sample, and 
 verifying the second model based on a first relationship between the first output data and the second output data and a second relationship between the third output data and the fourth output data. 
 
     
     
         8 . The method of  claim 7 , wherein the training the second model comprises
 training the second model such that a loss proportional to an error between the first relationship and the second relationship decreases.   
     
     
         9 . The method of  claim 1 , further comprising:
 adjusting at least one sub-process based on the estimated structure; and   manufacturing an integrated circuit through a semiconductor process comprising the adjusted at least one sub-process.   
     
     
         10 . A system comprising:
 at least one processor; and   a non-transitory storage medium storing instructions which, when executed by the at least one processor, instruct the at least one processor to perform measurement of a structure based on a spectrum of the structure,   wherein the measurement of the structure comprises
 obtaining a first model trained based on simulation data, the first model including a first sub-model and a second sub-model following the first sub-model; 
 generating a second model such that the second model includes a third sub-model generated from at least a portion of the first sub-model; 
 training the second model based on sample spectrum data generated by measuring spectra of sample structures; and 
 estimating, based on the trained second model, the structure from measured spectrum data generated by measuring the spectrum of the structure. 
   
     
     
         11 . The system of  claim 10 , wherein the obtaining the first model comprises
 generating virtual spectra by simulating virtual structures, and   wherein the simulation data represents the virtual structures and the virtual spectra.   
     
     
         12 . The system of  claim 11 , wherein the obtaining the first model comprises
 verifying the first model based on an error between output data of the first model and the simulation data.   
     
     
         13 . The system of  claim 10 , wherein
 the second model further comprises a fourth sub-model following the third sub-model, and   the training the second model comprises fixing the third sub-model such that the third sub-model is not trained while training the fourth sub-model based on the sample spectrum data.   
     
     
         14 . The system of  claim 13 , wherein
 each of the first sub-model and the third sub-model comprises a convolution network, and   each of the second sub-model and the fourth sub-model comprises a fully connected network.   
     
     
         15 . The system of  claim 10 , wherein the measurement of the structure further comprises
 verifying the second model based on an error between output data of the second model and measured structure data of the sample structures.   
     
     
         16 . The system of  claim 15 , wherein the verifying the second model comprises
 extracting a first sample and a second sample from the simulation data,   obtaining first output data and second output data of the first model, the first output data and the second output data respectively corresponding to the first sample and the second sample,   obtaining third output data and fourth output data of the second model, the third output data and the fourth output data respectively corresponding to the first sample and the second sample, and   verifying the second model based on a first relationship between the first output data and the second output data and a second relationship between the third output data and the fourth output data.   
     
     
         17 . The system of  claim 16 , wherein the training the second model comprises training the second model such that a loss proportional to an error between the first relationship and the second relationship decreases. 
     
     
         18 .- 25 . (canceled) 
     
     
         26 . A method for measuring a structure based on a spectrum of the structure, the method comprising:
 obtaining a first model trained based on simulation data, the first model including a first sub-model and a second sub-model following the first sub-model;   generating a second model based on the first model;   training the second model based on sample spectrum data generated by measuring spectra of sample structures;   verifying the trained second model based on output data of the first model and output data of the trained second model; and   estimating, based on the verified second model, the structure from measured spectrum data generated by measuring the spectrum of the structure.   
     
     
         27 . The method of  claim 26 , wherein the verifying the second model comprises:
 extracting a first sample and a second sample from the simulation data;   obtaining first output data and second output data of the first model, the first output data and the second output data respectively corresponding to the first sample and the second sample;   obtaining third output data and fourth output data of the second model, the third output data and the fourth output data respectively corresponding to the first sample and the second sample; and   verifying the second model based on a first relationship between the first output data and the second output data and a second relationship between the third output data and the fourth output data.   
     
     
         28 . The method of  claim 27 , wherein the training the second model comprises training the second model such that a loss proportional to an error between the first relationship and the second relationship decreases. 
     
     
         29 . (canceled)